NVIDIA shows signs ... [2008 - 2017]

"Neck and neck"? So it's only 30% faster than a GTX285? Even the 5800Ultra was a lot faster than the Ti4600...

No, not really. But GF100 and Cypress are both nearly 20% bigger than their preprocessors.

 

I'm definitely not an expert either. But, according to the 10Q, they do. The numbers add up exactly. And you have to admit: the alternative wouldn't make sense at all. Increase and decrease R&D expenses bij $90M just for 1 quarter?

 

Looks like they felt the need to update the article, still claiming that the y-o-y reduction is significant and that R&D should go up dramatically when new silicon needs to go out.

I've personally never experienced that: when getting closer and closer to tape-out, companies simply start to put more pressure on employees and social life may go away completely, but that's about it. It's not as if much more can be done anyway (see also: Brooks's Law).

Wrt to the y-o-y reduction in the last quarter, according to the 10Q:

Going from $212M to $197M, that leaves a whopping $6M in real y-o-y R&D reduction, further explained in the 10Q: basically good old fashioned cost control during lean times. (AMD, HP and my company can't be the only one where salaries, bonuses and other perks have been cut across the board.)

The article also ignores an increase of $5M between last quarter and the quarter before, coincidentally exactly the quarter where these costly tape-outs could have been expected. Bringing new chips up to production doesn't increase R&D costs dramatically, unless you're a startup that needs to buy/lease tons of lab equipment. Yes, there's tape-out costs, but that's by far the biggest part and the $5M more than compensates for that.

In his forum, Charlie brings up additional costs, such as additional compute costs (when you're close to taping out???) and overtime payment (I don't know if I have to laugh or cry.)

The 'journalist' of that article must have been paid real money for this...

 

http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MzIxMDB8Q2hpbGRJRD0tMXxUeXBlPTM=&t=1
NVIDIA posts fiscal Q4 2010 profit of $131.1 million

 

Notes:

• Revenue was nearly a billion dollars, gross margin nearly 45%.
• Nearly 100% growth quarter-on-quarter compared to the same period a year ago, 22% up on Q3 '09.
• They've got just about $1.3B in cash and sec lying around.
• Forecasts for Q1 '10 are flat, gross margin around 45% again.
• Jen-Hsun said they could have done another couple of hundred million dollars of business in that quarter if they hadn't been capacity constrained.

I have no frame of reference without checking, but it seems healthy enough. Lots of focus on Fermi and it shipping soon, and some fairly wooly stuff around Tegra and professional.

 

He didn't say anything upbeat about Geforce at all that I remember. He did talk about Fermi a lot but he sounded a lot more excited about its prospects in the Quadro and Tesla businesses.

 

Typo, or do they really label their Quarterly results 1 year ahead? Financials for Q4 2010, but for the quarter ending Jan 31, 2010? So now they are starting on Q1 2011?

Anyway, not bad. I expected a Q to Q loss in revenue, but they managed an 8.8% increase in revenual.

They managed to mitigate to some extent the damage of not having a competing part, however it still allowed the competition to make some large gains (reference 40% Q to Q gain in revenue for AMD GPU revenue).

Their guidance for Q1 2011 is flat. So they aren't expecting Fermi to have a large impact, obviously, as it'll only be on the market for a bit over 1 month.

Regards,
SB

 

Their R&D expenses for the quarter came in at $216M, second highest ever. I wonder how Charlie is going to spin that.

 

probably the same way that many here spin everything else in the slight chance to bash ChuckyD ..remember Charlie didn't write the article concerning the R&D spending, it was Luis Silva, yet Charlie bashers continue to recycle that hoping that maybe if they repeat it enough times maybe it will become true. Also Charlie himself @ S|A attributed several other costs that would have a put NV in a much better light (OMG.. yes really).. posting about previous spending over the last 2+ years that led to the article being updated.. DavidB, followed by Chocolate Factory, Azazel and P4man made many such comments well into the 2nd page of discussion let to it being updated accordingly.

For as much ire as Charlie draws it seems the fanbois really go out of their way in bringing him into the discussion whenever they can.

As for the quarterly earnings, it's in the black thanks in part to a lot of cost cutting over the last year (and as part of that a significant reduction in inventories). Also of note, iirc NV stated previously that margins in the professional markets (Quadro/Tesla) had gone up considerably which too would attribute to the satisfactory earnings. For discrete the real plus seems to be the GT21x series which saw an uptick in OEM use. I haven't taken a look at ATI's numbers but I'm willing to bet their earnings increased at a greater percentage in comparison across Q3-Q1. With Redwood just making its way into OEM and no updated (other than renames) answer from NV Q2 looks to be promising unless NV surprises everyone and fermi derivatives come hard and fast.

 

Would you not say that its been TSMC which mitigated ATIs advantage rather than say Nvidia themselves?

 

It's hard to quantify that as it affected Nvidia to some extent also. If rumors are to be believed that one of the main reasons for the Fermi delay is to get improved yields on Fermi.

But yes, I'd attribute some of that to TSMC's 40 nm woes, no doubt there would have been a lot more Cypress chips had the process been better.

Regards,
SB

 

Its kind of one of those 'what if' scenarios where the more you consider it the more complete the picture of how complicated the relationship really is.

 

I think that GF100's die is too big considering the relative immaturity of TSMC's 40 nm process. The last reticle sized GPU made ny was GT200 and it was on 55nm, an older at the time process.

Even lrb is unlikely to be manufactured at the leading process at intel as it's die size is too much.

 

http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MzIxMDB8Q2hpbGRJRD0tMXxUeXBlPTM=&t=1

They made 130mln profit in Q4 only, yet net loss for the year is 67mlns after taxes et all ?!

 

Sure, by loosing $201M in first and $105M in second quarter..

 

Nvidia shares dip 9% on inventory buildup
http://sanjose.bizjournals.com/sanjose/stories/2010/02/15/daily82.html?ana=yfcpc

So they increase inventory by almost 20% from the highest profit quarter of the year (Q4) to the lowest (Q1).
So they shipped the cards but the cards weren't sold to costumers because they don't want them but nvidia ship them anyway. They are building a bubble.

On the other side ATI didn't have enought cards to supply so much demand so let's see how is going to be next 2 quarters.

 

You must have overlooked this part in the conference call transcript:

 

What are they shipping? And who are they shipping to?

 

I wonder if its got something to do with the 40nm production they used late last year for their low end 40nm chips? They could have done this in order to deny ATI the wafers they needed for higher production last year without actually having orders for those chips to fulfill.

 

They're shipping crapload of 40nm GT2xx lowend chips, which at the moment are nothing but filling countles shelves on countless stores
They sold a lot of them too, and they're the sole reason nV managed to raise their market share, but regardless of that there's tons of them just lying on the shelves - from nVidia's point of view they're sold products, but in the end they're not sold.